HEBO suggests the same parameters multiple times

[qo4on]

Why does opt.suggest suggest the same parameters multiple times?

space_cfg = [
    {'name': 'n_estimators',     'type': 'int', 'lb': 20, 'ub': 21},
    {'name': 'max_depth',        'type': 'int', 'lb': 2, 'ub': 3},
]

{'n_estimators': 20, 'max_depth': 2}
{'n_estimators': 20, 'max_depth': 2}
{'n_estimators': 21, 'max_depth': 2}
{'n_estimators': 20, 'max_depth': 3}
The value argument must be within the support

HEBO suggests {'n_estimators': 20, 'max_depth': 2} twice and throws an error while there is one hp configuration remains untested: {'n_estimators': 21, 'max_depth': 3}.

[Alaya-in-Matrix]

It seems that the search space is so small (2 * 2 = 4), so that HEBO has enumerated all possible combination

[qo4on]

I believe this is incorrect behavior. @smcdonagh, @iamhankai please take a look at this.

[smcdonagh]

Hi @qo4on -- I am not part of the HEBO project, apologies.

[alexanderimanicowenrivers]

@qo4on This is correct behaviour and I will close this issue. If you look at your design space you only have two variables for n_estimators (20, 21) and two for max_depth (2,3) so there's going to be expected overlap, typical search spaces are much larger.